[Forum SIS] Avviso di Seminario prof. Y. Ritov

[Carlo Gaetan]

=========================================================
AVVISO DI SEMINARIO
=========================================================

Venerdi' 23 aprile 2010, alle ore 14.30
presso il Dipartimento di Statistica
dell'Universita' Ca' Foscari di Venezia,
Sala del Consiglio (secondo piano)
(http://www.unive.it/nqcontent.cfm?a_id=48021)


YA'ACOV RITOV (Department of Statistics, The Hebrew University of
Jerusalem)


terrà un seminario dal titolo:

Should we believe in our priors? Some examples of Bayesian
analysis of complex models.

Abstract

We will discuss three examples coming from different statistical fields.
The unified motive of the three is that a "natural" Bayesian analysis  may
be misleading. The first example is of a simple semiparametric  model.
There is a simple estimate for the parameter of interest (the
Horvitz-Thompson estimator). the Bayesian approach, however, enforces us
to ignore the crucial information needed for the estimator. Then we'll
consider hidden Markov model (HMM), in which the maximum-a priori (MAP)
estimator is trivial and false. Finally, we will consider a sparse
regression model in which the standard estimator can be considered as
MAP, but with prior that must change with sample size, and must imply
more smoothness than believed.



Tutti gli interessati sono invitati a partecipare.



-- 
Carlo Gaetan

Dipartimento di Statistica
Università Ca' Foscari - Venezia
San Giobbe, Cannaregio 873,
I-30121 Venezia
ITALY

phone: ++39  041 234 7411 (gen. office - Venice)
       ++39  041 234 7437 (desk - Venice)
       ++39 0422  51 3700 (gen. office - Treviso)
       ++39 0422  51 3747 (desk - Treviso)

fax  : ++39  041 234 7444 (Venice)
       ++39 0422  51 3715 (Treviso)

e-mail:[gaetan"at"unive"dot"it]
web:[http://www.dst.unive.it/~gaetan]

Please don't print this e-mail unless you really need to.
Please avoid sending me Word, Excel or PowerPoint attachments. See
http://www.gnu.org/philosophy/no-word-attachments.html.

Per favore non stampare questo messaggio se non è proprio necessario.
Per favore non mandatemi allegati in Word, Excel o PowerPoint. Le
ragioni sono spiegate qui
http://www.gnu.org/philosophy/no-word-attachments.it.html